COHERENCE - Cooking: leaderboard

COHERENCE (2026) removes the images from 6,161 interleaved image-text documents (WikiHow, StoryBird, cooking recipes, science) and asks the model to put each candidate image back in its place. This board is the exact-match accuracy on the cooking recipe documents, in percent; higher is better.

Metric: Exact-match accuracy (%) on the 1,326 cooking recipe documents; 6,161 interleaved image-text documents whose images were removed: the model assigns each candidate image to its placeholder; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 19 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1GPT-5.4High79.19#135
2Gemini 3.1 Pro (Preview)73.45#99
3Claude Sonnet 4.6Thinking73.15#173
4Qwen 3.5 397B A17B71.27#264
5Doubao-Seed-2.0-Pro-26021563.27#158
6Qwen 3.5 122B A10B61.54#346
7Kimi K2.558.97#266
8GPT-5.255.28#185
9Qwen 3.5 35B A3B54.15#492
10Intern-S1-Pro52.34#585
11Qwen 3 VL 30B A3B52.26#784
12Doubao-Seed-2.0-Lite-26021551.58#262
13Qwen 3 VL 235B A22B46.83#485
14Doubao-Seed-2.0-Mini-26021543.89
15Qwen 3.5 4B43.14#1241

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=coherence-cooking · How It Works · Data refreshed daily, snapshot 2026-11-21.